Research paper

Purpose

To verify whether an evolutionary model outperforms logistic regression in determining the institutional placement decisions made by a London social service department panel.

Design/methodology/approach 

Genetic chromodynamics models an algorithm within the Michigan evolutionary classifier. Hence multiple classification rules evolve simultaneously. The data set as described by (Xie et al., 2002) is used. Two thirds of randomly selected cases are for training and one third for testing. Indicator weights are set between 0 and 1.

Findings

Of 275 placements, 40% represent residential homes, 48% nursing homes, 12% nursing long-stay and two hospital long-stay. In ten runs, 89.18% were correctly placed (range 81.6 to 97.7%); 5.07% wrongly placed (range 1.2 to 8.0%) and 5.75% unplaced (range 0.0 to 11.5%). Changing the 0.99 weights to 0.90 and 0.80 placed 87.6% and 87.9% correctly. 

Research limitations/implications 

Data came from written records. Errors in transcription and placement could not be checked. Other facts, or the weights, may be influencing placement decisions. 
Practical implications
(Xie et al., 2002) matched 78% of 195 placements. The evolutionary model outperformed logistic regression both in placements evaluated (275/195) and accuracy (89%/78%). Therefore, it could be used as a first line management information tool, revealing whether guidelines are followed. 

Originality/value 

We have developed and tested a computational model, which could be used to evaluate institutional placement decisions in the UK ‘market’. Further development and exploitation would facilitate greater understanding of the needs old people and the resources necessary for their appropriate management.